In a moving image processing device, conventionally, moving image data of television programs is recorded on a recording medium, and the recorded data is read out and the moving image is reproduced. There is further known a technology in which in order that a user of the moving image processing device may easily recognize contents of moving images of recorded programs, a thumbnail image of static images generated based on the moving images is presented to the user (see, e.g., Japanese Laid-open Patent Publication No. 2006-339718, Japanese Laid-open Patent Publication No. 2007-281945, Japanese Laid-open Patent Publication No. 2003-32581, International Publication Pamphlet No. WO 2008-081514).
In addition, there is known a technology in which a feature amount of both of images is calculated and compared to search similar images (see, e.g., Japanese Laid-open Patent Publication No. 2004-341940).
However, there is a problem that a thumbnail image automatically generated based on moving images does not necessarily indicate contents of programs appropriately. For example, a method for extracting a thumbnail image from a predetermined portion of a moving image is considered by using as a standard a CM (commercial message) as in a technology disclosed in Patent Literature 4. However, a thumbnail image may be extracted without relation to content of a real moving image. As described above, it is difficult to automatically generate a thumbnail image so that a user may grasp contents of a program.